Mathematica Cookbook

Chapter 16. Parallel Mathematica

splintered dreams of unity (our lives are parallel) so far from reality (our lives are parallel) independent trajectories (our lives are parallel) separate terms of equality (our lives are parallel)

our lives are parallel

is there no redemption? no common good? is there nothing we can do for ourselves? or only what we should? comes the hard admission of what we don’t provide goes the insistence on the ways and means that so divide

—Bad Religion, “Parallel”

16.0 Introduction

Mathematica has impressive performance on many types of problems. The majority of Mathematica users are not drawn to Mathematica for its brute speed, but rather for its unparalleled depth of features in the realm of symbolic processing. Yet, there are certainly problems that you will solve in Mathematica that you will want to scale to larger data sets or more complex models. In the past, the only viable solution might be to port your Mathematica solution to C or Fortran. Today relatively cheap multiprocessor and multicore computers have become commonplace. My primary development machine has eight cores available. Wolfram provides two solutions for exploiting multiple CPUs. The first solution, called Grid Mathematica, has been available as a separate (and somewhat costly) product distinct from your vanilla Mathematica product. The second solution is available to everyone who has updated to Mathematica 7. One of the big feature enhancements in version 7 is integrated parallelism ...